Abstract
In low-exposure English as a Foreign Language (EFL) contexts, where only one organized session is available each week, generative artificial intelligence (AI) tools are increasingly used to supplement limited class time. Few studies have explored how teachers and students within the same non-formal institution use these technologies. Non-formal settings remain under-researched, such as the Taman Pendidikan Al-Qur'an (TPQ), a non-formal Islamic educational institution in Bandung Regency, West Java, Indonesia, that teaches English once a week. A qualitative descriptive study assessed the perceptions of 17 students and 5 teachers regarding AI. Data from a bilingual, open-ended reflective questionnaire were thematically analyzed in NVivo 14 across five domains: awareness, perceived benefits, challenges, fears and concerns, and motivation for future use. The findings indicate that the divergence is role-based, not generation- or attitude-based. Students were impressed by how quickly AI could help with translation and grammar, and they could access it outside class. Teachers were more cautious, pointing to cognitive dependence, output accuracy, and the loss of independent learning. Both groups expressed ethical concerns, but about different forms of over-reliance. The gap is attributable not to incompatible mindsets about technology but to structural issues: limited contact hours, uneven digital proficiency, and the lack of institutional AI policy. These results suggest that AI literacy interventions in non-formal EFL contexts should involve teachers and students together and that national AI-in-education frame.